An Experimental Study of Autoignitionin Turbulent Co-Flows of Heated Air

C.N. Markides & E. Mastorakos

Hopkinson Laboratory, Department of Engineering,University of Cambridge, U.K.

INTRODUCTION• Theory:

Motivated by the DNS work of Mastorakos et al, 1997(and similar)– Re-examination of laminar, inhomogeneous Linan, Linan/Crespo, mid-70’s– Maximizing local reaction rate through ξMR (most reactive mixture fraction)

– AND –– Minimizing local heat losses through χ (effect of scalar dissipation rate)– “Turbulence” may accelerate autoignition– Autoignition was always observed at a finite τIGN (ignition delay time)

• Experiment:Turbulent, inhomogeneous counterflows of Law et al, from late-90’s(and similar)– Turbulent, hot air opposite cold fuel, including hydrogen (elliptic problem)– Enhanced turbulence and increased strain rate increase “autoignition

temperature” necessary for autoignition – and even more interestingly –– Higher strain rates completely preclude autoignition

OBJECTIVES• Aforementioned results are not entirely consistent and there is

an inability to properly explain why

• This is a reflection of a more general situation:– Insufficient current knowledge concerning turbulent,

inhomogeneous autoignition– Limited number of relevant, well characterized experiments for

validation– THUS –

• In order to understand the fundamental underlying physics of the coupling between turbulent mixing and the chemistry of autoignition, we experimentally:– Observe autoignition in a turbulent, co-flow configuration

(parabolic problem, easier to model)– Investigate the temporal and topological features of the

phenomenon– Results directly available for modelling

APPARATUS• Air continuously through

Perforated Grid (3mm, 44%) & Insulated Quartz Tube (24.9mm):– Velocity: up to 40m/s– Temperature: up to 1200K– Turbulence Intensity: 12–14%– Integral Length-scale: 3–4mm– Returb: 80 - 220

• Atmospheric Pressure• Fuel continuously through

S/Steel Injector (2.24/1.185mm):– Velocity(*): 20–120m/s– Temperature(*): 650–1000K– Limited control of temperature

• Bluff bodies (10.0 & 14.0mm):– Used with 24.8 & 34.0 mm tubes

to give a single blockage ratio 0.17

INDEPENDENT VARIABLES:EXPERIMENTAL ACCURACY

• Set all rates to get a steady and repeatable flow– AIR- and FUEL-MFC (excellent, <0.6%)– N2 Flow Meter (average, <5%)

• Measure all flow rates accurately– AIR- and FUEL-MFC (excellent, ~0.9% and ~1.9%)– N2 Flow Meter (average, <6%)

• Set heaters to get steady temperature conditions– Active Heater Controllers (excellent, <1K)

• Measure Tair and Tfuel accurately– Air stream (excellent, <4K(random)+6K(systematic), or <1%)– N2-diluted fuel stream (good, <14K+2K, or <2-3%)– N2-diluted fuel stream & small injector (average, <14K+12K, or <3-

4%)• Measure geometry accurately

– Quartz tubes (excellent, <0.03mm or <0.1%)– Normal injectors (good, <0.03mm or <1%)– Small injectors (excellent, <0.005mm, or <0.4%)

• Measure the ambient pressure• Use accurate 2nd Order Virial Equation of State (error<1%) for

densities

INDEPENDENT VARIABLES:CHARACTERIZATION

• PITOT TUBE and HOT WIRE– Profiles at various axial locations for different Returb

– Mean velocity field uniformity– Magnitude of turbulence intensity– Integral lengthscale from Taylor hypothesis– Turbulence spectra estimation– Kolmogorov scales (dissipation) from variance of the velocity spatial

gradients

• THERMOCOUPLE– Profiles at various axial locations– Heat losses– Extent of thermal boundary layer (profile uniformity)– Estimate temperature fluctuations

• HIGH TEMPERATURE HOT WIRE– Attempt to get simultaneous fluctuations of temperature and

temperature/velocity fluctuation cross-correlations

BULK BEHAVIOUR• CTHC: Four regimes of operation identified for

given Yfuel:

1. ‘No Ignition’2. ‘RANDOM SPOTS’3. ‘Flashback’4. ‘Lifted Flame’

• CTHAJ: Similar, with exception of ‘SPOT-WAKE INTERACTIONS’

T

U

RandomSpots

Flashback

NoIgnition

LiftedFlame

• Looking at effects of:– Fluid mechanics

• Uair and Ufuel

– Chemistry• Tair and Tfuel(*)• Fuel dilution with N2 (Yfuel)

Flow Direction

Injector

Quartz Tube

UNSTEADY BEHAVIOUR

0 10 20 30 40 50 600

50

100

150

Time (s)

Min

imum

Aut

oign

itio

n L

engt

h (m

m) Spot-Wake Interaction

Random Spots

• CTHAJ:‘Spot-Wake Interactions’

• Velocity/Mixing PDFs crucial

• CTHC:‘Unsteady Regime’?

• Velocity/Mixing PDFs crucial

0 10 20 30 40 50 600

20

40

60

80

100

120

Time (s)

Min

imum

Aut

oign

itio

n L

engt

h - L

MIN

(mm

)

Unsteady Regime: Uair

=Ufuel

=24.8m/s,Tair

=877K,YC2H2

=0.66U

air=U

fuel=25.4m/s,T

air=902K,Y

C2H2=0.55

Uair

=Ufuel

=24.8m/s,Tair

=874K,YC2H2

=0.53

Confined Turbulent Flows of Hot AirConfined Turbulent Flows of Hot Air

Quartz Tube:24.9mm

Insulation: Blanket,‘Jacketed’ Tube,Heat Exchanger

Injectors: 2.24&1.185mm

Fuels: H2,C2H2, C2H4,

n-C7H16

Mixing w/ Acetone PLIF and Link w/ LIGN

MEASURE:LIGN, τIGN and fIGN

Quartz Tubes:24.9&34.0mm

Insulation:Blanket,

‘Jacketed’ Tube

Injector &Bluff-bodies:

2.24&10.0/14.0mm

Fuels:

C2H4

only

MEASURE:LIGN ONLY and fIGN

Confined Turbulent Hot Co-FlowsConfined Turbulent Hot Annular Jets

REVIEW

OPTICAL MEASUREMENTS – I SPECTROSCOPY

• CTHC and CTHAJ similar

1. Nothing-to-Spots Transition: C2H4

2. Random Spots: H2

3. Comparison: C2H2 and H2

1

2 3

OPTICAL MEASUREMENTS – II IMAGING

Injector

2.5 mm

~ 4 mm ø

Flow Direction

Flow Direction

Flow Direction

OPTICAL MEASUREMENTS – IIIPMT

• Fast imaging and PMT with all fuels including H2

• Reveal characteristic autoignition event profiles:explosion, propagation and quench

• Obtained fIGN from PMT timeseries; strong correlations with LIGN

OPTICAL MEASUREMENTSOVERVIEW

• Post-ignition flamelet propagation images consistent with DNS

– Spherical shell shape– Propagation velocities ~ 15–20m/s for C2H2 (not considered

in depth)

• Life-span of spots ~ 0.1–0.2s for C2H2 but can vary across fuels

• Autoignition kernel propagation velocities ~ Uair

• Exposure times important because they determine theautoignition information that can be retrieved from the raw images

Flow directionEarliest

Mean

IMAGING DATA ANALYSIS

Earliest

Mean

• Lower U (~ 20 m/s)• And/or Higher T (~ 1010

K)

• Higher U (~ 26 m/s)• And/or Lower T (~ 1000

K)

PDFs from“OH Snapshots”

• From PDF image get lengths:– Mean L⟨ IGN. and ⟩ Standard Deviation LRMS

– Earliest LMIN

• Attempt to define corresponding times

LMIN

⟨LIGN.⟩

Flow Direction

PRELUDE TO RESULTS• In-homogenous autoignition of fuels

in a turbulent co-flow of hot airwith/without an additional bluff-body

• Various regimes possible, depending on conditions– We concentrate on the ‘Random Spots’

• Three types of experiments (mixing):– Equal velocities in CTHC– Jet in Co-Flow in CTHC– Jet in CTHAJ

• (Mostly) optical OH chemiluminescence measurements (images)– To get PDF of autoignition– Define suitable “autoignition lengths”– And calculate corresponding “residence times until autoignition” or

“autoignition delay times”

CTHC RESULTS – I (H2)• Lengths:

– Equal Velocity Case (Uair = Ufuel)

– Increased Tair shifts autoignition UPSTREAM

– Increased U shifts autoignition DOWNSTREAM• LMIN ~ 60–70% of L⟨ IGN⟩

• Times:– Define τMIN “minimum autoignition time” simply as: LMIN/U (~ 1 ms)

– Increased Tair → EARLIER autoignition

– Increased U → DELAYED autoignition

• Similarly for Jet in Co-Flow:– Not easy to define an unambiguous “autoignition time”– Consider the centreline velocity decay in the jet and integrate

IncreasingTair

IncreasingU

U

T

Increasing U

T

U

840 850 860 870 880 890 900 910 9200

20

40

60

80

100

120

Air Temperature (K)

LM

IN,

LM

IN

and L

IGN

(m

m)

U=17.5,Y=0.60 U=20.7,Y=0.62U=29.8,Y=0.62

10 12 14 16 18 200

1

2

3

4

5

6

7

8

9

10

Uair

(m/s)

MIN

(m

s)

T=1026,Y=0.73T=1043,Y=0.72

CTHC RESULTS – II (HnCm)

0.4 0.5 0.6 0.7 0.8 0.9 10

20

40

60

80

100

120

140

Yfuel

(-)

LM

IN a

nd

LM

IN

(mm

)

T=821,U=11.2T=876,U=24.8T=902,U=25.5

T=1026,U=14.8T=1042,U=18.3

1. Effect of fuel dilution (C2H2&C2H4 ):– LIGN decreases as Yfuel increases

2. Effect of Uair (C2H4):– τIGN increases as Uair increases

3. Effect of Tair and small injector (C2H2):– LIGN decreases as Tair increases

– Sensitivity of Tair lost for small injector

• On the effect of Uair:– Autoignition delayed by increase in Uair (and hence) u’,

(because u’ increases with Uair so that u’/U ~ const. behind the

grid)

– BUT –

– Direct comparison with DNS pre-mature until ξ and χ measurements are considered

– In other words:u’ increases, but does χ ~ u’/Lturbξ’’2 also locally increase?

PRELIMINARY DISCUSSION

TURBULENT MIXING:ΒACKGROUND

• Acetone PLIF for mixture fraction

• 266nm straight form Nd:Yag, 110mJ/pulse– Sheet thickness <0.1mm (Kolmogorov Length scales are >

0.15mm)

• Optimal linear de-noising (Wiener) of all images in the Wavelet domain before taking gradients for χ2D

• Consider justification for extending to χ3D

• We have <ξ>, <ξ‘2>, <χ>, <χ‘2> and (not shown) pdf(ξ), pdf(χ)

– Also conditional <χ|ξ>, <χ|ξ‘2>, pdf(χ|ξ)

TURBULENT MIXING: <ξ>

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

0.8

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

0.8

0.2

0.4

0.6

0.8

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

-2 0 20

5

10

15

20

r/d (-)

z/d

(-)

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

-2 0 20

5

10

15

20

r/d (-)

z/d

(-)

• BELOW:– All are equal velocity cases (Uair =

Ufuel) with varying Returb

• RIGHT:– Jet case (Ufuel = 3 and 4 Uair)

0 5 10 15 20 250

0.5

1

z/d (-)

(-

)

10-2

100

10210

-4

10-2

100

z/d (-)

(-

) -2 z/d=5

TURBULENT MIXING (Uair=Ufuel):<ξ>

-2 -1 0 1 20

0.5

1

r/d (-)

(-

)

z/d=1

z/d=4 z/d=5

z/d=6

0 0.05 0.1 0.1510

-1

100

(r/d)2/(z/d)2 (-)

/ (

r=0)

(-

)

z/d=1

z/d=4, 5 and 6

TURBULENT MIXING (Uair=Ufuel):<ξ'2>

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

0.02

0.06

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

0.02

-2 0 20

2

4

6

8

10

r/d (-)

z/d

(-)

0 5 10 150

0.05

0.1

0.15

z/d (-)

'

2 (

-)

Re=404, =0.93Re=532, =0.94Re=561, =0.97Re=561, =1.85Re=746, =1.89

TURBULENT MIXING (Uair=Ufuel):<χ>

-1 0 10

1

2

3

4

5

r/d (-)

z/d

(-)

10

20

30

40

50

60

70

-1 0 10

1

2

3

4

5

r/d (-)

z/d

(-) 0 5 10 15 20

0

50

100

z/d (-)

2D

(

1/s)

Re=404, =0.93Re=532, =0.94Re=561, =0.97Re=561, =1.85Re=746, =1.89

-1 0 10

50

100

150

200

r/d (-)

2D

(

1/s)

0 10 20 3002

5

10

15

z/d (-)

3D

/

'2 .

turb

TURBULENT MIXING (Uair=Ufuel):MODELLING – Isotropy and CD

10-3

10-2

10-1

100

101

10210

-3

10-2

10-1

100

101

102

axial

(1/s)

ra

dia

l (

1/s

)

• LEFT:– Isotropy (Radial and Axial Components of χ2D)

• RIGHT:– Timescale ratio model for <χ2D> only valid away from the

injector

CONCLUSIONS

• Length (both LMIN and L⟨ IGN.⟩):– Increase non-linearly with lower Tair and/or higher Uair

– Increase with Ufuel

• Residence Time until Autoignition:– Increases with lower Tair and/or higher Uair

• Enhanced turbulent mixing through u’ and through <χ>:DELAY AUTOIGNITION

An Experimental Study of Hydrogen Autoignition

in a Turbulent Co-Flow of Heated Air

C.N. Markides & E. Mastorakos

Hopkinson Laboratory, Department of Engineering,University of Cambridge, U.K.